Hi Anton,
reading from database should be faster than checking at the file system, that's why we took this approach. Also, database allows us to store more data about the file like the original URL of the file, its timemodified and revision in Moodle, its download time, etc. Some of this data could be calculated if we read the file system, but for example we wouldn't be able to tell if the file needs to be updated or not because we don't know the time it was modified in Moodle.
In any case, I'll write down your suggestion so we can take a look at it in the future, but you shouldn't expect a change soon.
Cheers,
Dani